Skip to content

Latest commit

 

History

History
169 lines (127 loc) · 7.73 KB

FAQ.md

File metadata and controls

169 lines (127 loc) · 7.73 KB

Frequently Asked Questions

How do I interact with my docker images?

To check what docker images you have built, run docker image ls, below is example output.

REPOSITORY                    TAG            IMAGE ID       CREATED        SIZE
nyuappsec/assign3-db          v0             18756add02dd   2 days ago     444MB
nyuappsec/assign3-proxy       v0             bab62e6f1d2d   2 days ago     47.6MB
nyuappsec/assign3             v0             51c1d0517639   2 days ago     379MB

What if I want to delete a Docker image? run docker image rm <REPOSITORY>/<TAG>. For example, if we want to delete nyuappsec/assign3:v0, run docker image rm nyuappsec/assign3:v0. You would see the following output,

Untagged: nyuappsec/assign3:v0
Deleted: sha256:51c1d05176395b10c4d2db6f7329fd9f099ef0f9ddca2865600ef5e67b03de73

# docker image ls
REPOSITORY                    TAG            IMAGE ID       CREATED        SIZE
nyuappsec/assign3-db          v0             18756add02dd   2 days ago     444MB
nyuappsec/assign3-proxy       v0             bab62e6f1d2d   2 days ago     47.6MB

How do I shut everything down and restart from scratch?

You can use kubectl delete -f to remove the resources specified in a YAML file. So for the resources in this assignment:

kubectl delete -f proxy/k8
kubectl delete -f GiftcardSite/k8
kubectl delete -f db/k8

Note that removing pods will not remove their persistent storage, and even deleting the persistent volume and persistent volume claim won't do it. To actually delete the data, you'll need to use minikube ssh to get access to the host container, and then remove the files there.

For example, to get rid of the persistent volume associated with the database (mysql-pv), you would do:

youruser@$ minikube ssh
docker@minikube:~$ sudo rm -rf /data/mysql-pv
docker@minikube:~$ logout

You can find out where the data for a persistent volume is located by looking at its hostPath in the PersistentVolume YAML file:

  hostPath:
    path: /data/mysql-pv

How do I see what's going wrong?

Start by looking at what pods are running, and their states:

kubectl get pods

If a pod is in a state other than Running, you can get more details about it:

kubectl describe pod <pod_name>

You can also ask for logs from a pod:

kubectl logs <pod_name>

Finally, it might be helpful to run a shell inside the pod's container so that you can run commands and see what's going on. You can do that by running:

kubectl exec --stdin --tty <pod_name> -- bash

Some pods (e.g., the GiftcardSite pod) don't have Bash installed, so you'll need to use sh instead.

Why can't kubectl find my Docker image?

The most common reason is that you forgot to run eval $(minikube docker-env) before running docker build. This points to the Docker client at the minikube Docker daemon.

Another common mistake is specifying the image version in the YAML file as latest (e.g., my_image:latest). This will make kubectl try to fetch the image from a remote repository, which will fail if it's an image you've created locally. You can override this behavior either by removing the :latest tag, or by adding imagePullPolicy: Never to the YAML file.

How can I open the web interface for a service?

You can find out what services are available by doing minikube service list:

caterina:k8s moyix$ minikube service list 
|----------------------|-------------------------------|--------------|-----|
|      NAMESPACE       |             NAME              | TARGET PORT  | URL |
|----------------------|-------------------------------|--------------|-----|
| default              | assignment3-django-service    |         8000 |     |
| default              | kubernetes                    | No node port |
| default              | mysql-service                 | No node port |
| default              | prometheus-alertmanager       | No node port |
| default              | prometheus-kube-state-metrics | No node port |
| default              | prometheus-node-exporter      | No node port |
| default              | prometheus-pushgateway        | No node port |
| default              | prometheus-server             | No node port |
| default              | proxy-service                 |         8080 |     |
| kube-system          | kube-dns                      | No node port |
| kubernetes-dashboard | dashboard-metrics-scraper     | No node port |
| kubernetes-dashboard | kubernetes-dashboard          | No node port |
|----------------------|-------------------------------|--------------|-----|

Then pick a service and use minikube service <name> to open a browser to that service's URL:

caterina:k8s moyix$ minikube service prometheus-server
|-----------|-------------------|-------------|--------------|
| NAMESPACE |       NAME        | TARGET PORT |     URL      |
|-----------|-------------------|-------------|--------------|
| default   | prometheus-server |             | No node port |
|-----------|-------------------|-------------|--------------|
😿  service default/prometheus-server has no node port
🏃  Starting tunnel for service prometheus-server.
|-----------|-------------------|-------------|------------------------|
| NAMESPACE |       NAME        | TARGET PORT |          URL           |
|-----------|-------------------|-------------|------------------------|
| default   | prometheus-server |             | http://127.0.0.1:61867 |
|-----------|-------------------|-------------|------------------------|
🎉  Opening service default/prometheus-server in default browser...
❗  Because you are using a Docker driver on darwin, the terminal needs to be open to run it.

How do I know if metrics are being scraped?

You can check the /metrics endpoint to verify any new targets, notice the database_error target in this image:

Metrics Endpoint

How do I know if I set up the proxy job correctly in configmaps?

The proxy-job entry in the Prometheus targets page represents a monitoring job configured to scrape metrics from a service running at http://proxy-service:8080/metrics. This image shows proxy-job is currently up, meaning Prometheus can successfully reach and retrieve metrics from it.

Proxy Job

How do I know if I have finished Part 2?

This image is a Prometheus graph displaying a metric named database_error_return_404_total. It represents the total count of 404 Not Found errors returned by a database over the past 12 hours. The graph's flat line for the majority of the time indicates a lack of these errors or a constant count. Notably, there's a sharp rise towards the end, signaling a sudden increase in 404 errors. This is because I manually caused the error by visiting a page that resulted in error 404, then after waiting a few moments for the scrape, I can view the increase in the graph. Graph

Here is a gif demonstrating the various actions and activities you can use Prometheus for. Prometheus Animation

Other things to note

  • We recommend using the docker driver for minikube (you'll see a message like "✨ Using the docker driver based on existing profile" when running minikube start). If you find that it's using some other driver, run two commands listed below. There's a bit more information in the documentation.
minikube delete
minikube start --driver=docker
  • When removing the mysql-related parts of the cluster, you might find that kubectl hangs waiting for the persistent volume claim to be deleted. A workaround is to run the following from another terminal window (or run it before trying to delete the mysql stuff).
kubectl patch pvc mysql-pvc -p '{"metadata":{"finalizers":null}}'